|Publication number||US7340552 B2|
|Application number||US 11/642,511|
|Publication date||Mar 4, 2008|
|Filing date||Dec 21, 2006|
|Priority date||Feb 18, 1992|
|Also published as||US5671371, US5881255, US5941973, US6128688, US6219738, US6341323, US6519667, US7177970, US20010023462, US20020032817, US20030070020, US20070106832|
|Publication number||11642511, 642511, US 7340552 B2, US 7340552B2, US-B2-7340552, US7340552 B2, US7340552B2|
|Inventors||Nobukazu Kondo, Seiji Kaneko, Koichi Okazawa, Hideaki Gemma, Tetsuya Mochida, Takehisa Hayashi|
|Original Assignee||Hitachi, Ltd.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (31), Non-Patent Citations (5), Referenced by (1), Classifications (14), Legal Events (1)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This is a divisional application of Ser. No. 10/274,881, filed Oct. 22, 2002, now U.S. Pat. No. 7,177,970 which is a continuation application of U.S. Ser. No. 09/991,925, filed Nov. 26, 2001, now U.S. Pat. No. 6,519,667; which is a continuation application of U.S. Ser. No. 09/777,960, filed Feb. 7, 2001, now U.S. Pat. No. 6,341,323 B2; which is a continuation application of U.S. Ser. No. 09/514,351, filed Feb. 28, 2000, now U.S. Pat. No. 6,219,738; which is a continuation application of U.S. Ser. No. 09/296,660, filed Apr. 23, 1999, now U.S. Pat. No. 6,128,688; which is a continuation of U.S. Ser. No. 09/203,621, filed Dec. 1, 1998, now U.S. Pat. No. 5,941,973; which is a continuation application of U.S. Ser. No. 08/847,974, filed Apr. 21, 1997, now U.S. Pat. No. 5,881,255; which is a continuation application of U.S. Ser. No. 08/544,727, filed Oct. 18, 1995, now U.S. Pat. No. 5,671,371; which is a continuation application of U.S. Ser. No. 08/016,692, filed Feb. 11, 1993, now abandoned.
The present invention relates to a bus control system for use in a data processing apparatuses such as a personal computer and a work station, and in particular, to improvement of a bus control system supporting a so-called split transfer protocol in which between a start cycle of an access operation of a processor and a response cycle for the access operation from an input/output (I/O) device related thereto, it is possible to insert on an identical bus a start cycle of an access operation of another processor.
As a bus like a conventional system bus, there has been used in many cases a bus supporting the split transfer protocol, for example, as described in “Futurebus+, P896.1, Logical Layer Specifications” (1990, IEEE). This is because that the utilization efficiency and the response time of the bus are improved.
On the other hand, the destination module designated by the address obtains the mastership of the bus ADDT when read data becomes ready for the access. The destination module then enables the signal ADRV and sends an address specifying a module to be accessed onto the bus ADDT. That is, it is to be noted that the same address is outputted onto the bus ADDT from the source and destination modules.
Simultaneously, the initiating module reports the terminating module via the line CONT that the access being initiated is a response to the split read access (at a timing 1302 of
The source module checks the contents on the line CONT and the access destination address on the bus ADDT to determine that the data is sent in response to the initiated access operation so as to get the response data.
However, as above, in a case where there is disposed a cycle in which the access destination address is outputted onto the bus ADDT when the response data is transferred in response to a split read access, the ratio of busy time of the bus in which the bus is being occupied for operation is increased. Recently, there has been an increase in the number of systems in which, also for minimization of the size and price, the number of signal lines of the bus is decreased, particularly, address and data lines are multiplexed in the bus. In such a multiplex bus, the increase in the busy ratio of bus is an essential problem because of deterioration in the bus utilization efficiency and increase in the response time.
Moreover, due to the recent growing volume of data to be processed, the number of address lines is also increased. In consequence, according to the method above, there exists a problem that the number of flip-flop circuits to keep therein addresses specifying access destination items is increased and hence the hardware system of each module becomes to be more complex.
It is therefore an object of the present invention to provide a bus control system capable of improving the utilization efficiency of the system bus and decreasing the response time to an access.
In order to achieve the object above, according to the present invention, each module connected to a bus is assigned with an identifier (ID) as identification thereof such that a module initiating an access operation outputs in a start or initiation cycle an address of the access destination onto the bus and an identifier of the initiating module onto a module identifier transfer line disposed as a separate line with respect to the bus, thereby notifying the address and the identifier to the module of the access destination. In response thereto, the destination module sends data onto the bus and an identifier of the initiating module onto the module identifier transfer line, thereby transmitting the data and the identifier to the initiating module.
Furthermore, even when the system includes a plurality of buses configured in a hierarchic structure, there is only a need to assign an identifier to each bus adapter (B/A) disposed between the buses to establish interface therebetween.
In addition, if necessary, an identifier may be similarly assigned to each module connected to the bus in each hierarchic layer. In this case, even when a plurality of modules connected to a hierarchic layer initiate access operations to modules connected to buses in other layers in a sequential manner with respect to time, the bus adapter related to the initiating modules can appropriately distribute response data items to the respective modules based on the identifiers thereof. Namely, in a multimedia system, each processor can output an I/O access onto an identical system bus in a concurrent fashion; consequently, the response time is minimized for an access request on the system bus.
These and other objects and advantages of the present invention will become apparent by reference to the following description and accompanying drawings wherein:
Referring now to
On the other hand, the destination module having received the split read access issues a request for the bus mastership when a read data item to be sent to the source module is ready for transmission. On acquiring the mastership, the destination module enables the data valid signal DATAV and outputs an effective read data item onto the bus ADDT[0-63]. Simultaneously, the destination module notifies that the transfer data being returned is a reply to the split read access to the source module via the line CONT (at a timing 106 of FIG. 1). Moreover, at the same time, the destination module outputs the identifier of source module via the line SINKMOD[0-3] to the source module (at a timing 104 of
Thereafter, the destination module releases the mastership of the bus ADDT to finish the response cycle.
The initiating module checks information sent from the destination module, namely, the mode specification control signal and the identifier of the source module on the line SINKMOD to determine an answer to the access issued therefrom. As a result, the initiating module obtains the response data.
Reference numerals 203 and 204 respectively designate bus adapters as interface units between the system bus 205 and the high-speed processor buses 201 and 202 to transfer data therebetween. Numerals 206 to 208 respectively denote bus adapters for establishing interface between the system bus 205 and the I/O buses 209 to 211 to transfer data therebetween. In this embodiment, the bus adapters 203, 204, 206, 207, and 208 are assigned with identifiers “0”, “1”, “2”, “3”, and “4”, respectively.
The embodiment shown in
In this construction, in accordance with the idea related to
The bus adapter 308 as the access destination module connected to the system bus 307 transmits the split read request from the source module to the I/O device 314. On receiving a response thereto from the I/O device 314, the bus adapter 308 sends data associated with the split read access from the initiating module to the system bus 307 corresponding to ADDT[0-63] shown in
Checking the identifier on the line SINKMOD [0-3], the initiating module 305 recognizes that data on the system bus 307 is response data of the split read access initiated by the module 305 and then causes the data to be sent onto the processor bus 304 so as to pass the data to the processor 301 having issued the read request.
In the diagram of
The bus adapter 305 includes an own ID register 5006 for keeping therein an identifier ID (“0” in the case of
The bus adapter 308 includes an own ID register 5016 for keeping therein an identifier ID (“0” in the case of
In this regard, reference numerals 5026 to 5029 stand for control lines, numeral 5030 indicates a control signal line of the system bus 307, a numeral 5031 is an identifier transfer line of the system bus 307, and numeral 5032 is an address/data line of the system bus 307.
Next, the operation of the bus adapter 305 will be described.
The bus adapter 305 simultaneously outputs an address for a read operation to the line 5032 and the value of the own ID register 5006 to the line 5031.
The bus adapter 308 invoked by the bus adapter 305 acquires the address and then initiates accessing an I/O device (the device 314 in the case of
Reading data from the I/O device via the I/O bus 311, the bus adapter 308 returns the data onto the line 5032 of the system bus 307. Simultaneously, the adapter 308 transmits the value of the source ID buffer to the line 5031.
After initiating the read operation, the adapter 305 causes the comparator 5009 to continuously compare the identifier on the line 5031 and the value of the own ID register 5006. Only in a data cycle when the identifiers match each other, the adapter 305 acquires the response data from the interface unit 5011.
As above, thanks to the construction shown in
In this regard, as can be seen from
In the system of
In the second embodiment shown in
In this case, the bus adapter 405 initiates, for the bus adapter 408, the split read accesses respectively of the processors P1, P2, and P3 in this order and sends at the same time the identifiers “0”, “1”, and “2” via the line SINKMOD[0-3] to the bus adapter 408.
The bus module 408 awaits, after accessing three I/O devices related thereto, responses from these I/O devices. Since the I/O device associated with the read request from the processor P3 sends the first response, the bus adapter 408 adds the source identifier “2” to the response data from the I/O device to send the resultant data to the system bus 407. Checking the identifier on the line SINKMOD of the system bus 407, the bus adapter 405 detects the source identifier “2” and recognizes that the identifier is assigned to the processor P3 related to the adapter 405, thereby passing the response data to the processor P3. The response data is transferred as indicated by a solid arrowheaded line in
Similarly, the next response data is appropriately sent to the processor P2 by the bus adapter 405 according to the value of the identifier “1” on the line SINKMOD. The flow of response data in this case is as denoted by a solid arrowheaded line in
In the similar manner, also the last response data is appropriately sent to the processor PI by the bus adapter 405 according to the value of the identifier “0” on the line SINKMOD. The flow of response data in this case is as designated by a solid arrowheaded line in
The bus adapter 405 accomplishing the operation above can be easily implemented by slightly modifying the bus adapter 305 or 308 of
This diagram is drawn on assumption as follows. A cycle 501 is a start cycle of a read operation, the bus adapter 405 has the bus mastership, and the initiating module (the response destination of the read data) is indicated as “0” (the processor 401 as the source module) on the line SINKMOD. A cycle 502 is a start cycle of a read operation, the bus adapter 405 has the bus mastership, and the initiating module is indicated as “1” (the processor 402 as the source module) on the line SINKMOD. A cycle 503 is a start cycle of a read operation, the bus adapter 405 has the bus mastership, and the initiating module is indicated as “2” (the processor 403 as the source module) on the line SINKMOD.
A cycle 504 is a response cycle of a read operation, the bus adapter 408 has the bus mastership, and the initiating module is indicated as “2” (the processor 403 as the source module) on the line SINKMOD. A cycle 505 is a response cycle of a read operation, the bus adapter 408 has the bus mastership, and the initiating module is indicated as “1” (the processor 402 as the source module) on the line SINKMOD. A cycle 506 is a response cycle of a read operation, the bus adapter 408 has the bus mastership, and the initiating module is indicated as “0” (the processor 401 as the source module) on the line SINKMOD.
There are shown in
Reference numerals 1001 and 1007 denote start cycles of I/O access of the processor P1, numerals 1002 and 1008 stand for response cycles of I/O access of the processor P1, numerals 1003 and 1009 designate start cycles of I/O access of the processor P2, numerals 1004 and 1010 stand for response cycles of I/O access of the processor P2, numerals 1005 and 1011 denote start cycles of I/O access of the processor P3, numerals 1006 and 1012 indicate response cycles of I/O access of the processor P3.
As can be seen from
Incidentally, since the main memory 406 is assigned with the identifier “3” as shown in
In this regard, according to the first and second embodiments, the identifier of the module initiating the split read access is transferred via the identifier transfer line SINKMOD. However, in the third embodiment shown in
While particular embodiments of the invention have been shown and described, it will be obvious to those skilled in the art that various changes and modifications may be made without departing from the present invention in its broader aspects.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US3997896||Jun 30, 1975||Dec 14, 1976||Honeywell Information Systems, Inc.||Data processing system providing split bus cycle operation|
|US4128883||Sep 30, 1977||Dec 5, 1978||Ncr Corporation||Shared busy means in a common bus environment|
|US4232366||Oct 25, 1978||Nov 4, 1980||Digital Equipment Corporation||Bus for a data processing system with overlapped sequences|
|US4281380||Dec 27, 1978||Jul 28, 1981||Harris Corporation||Bus collision avoidance system for distributed network data processing communications system|
|US4290102||Oct 25, 1978||Sep 15, 1981||Digital Equipment Corporation||Data processing system with read operation splitting|
|US4543628||Jan 28, 1983||Sep 24, 1985||Digital Equipment Corporation||Bus for data processing system with fault cycle operation|
|US4672662||May 15, 1985||Jun 9, 1987||Fujitsu Limited||Switching system and method for network having a plurality of terminal control equipment units|
|US4785394||Sep 19, 1986||Nov 15, 1988||Datapoint Corporation||Fair arbitration technique for a split transaction bus in a multiprocessor computer system|
|US4797815||Nov 22, 1985||Jan 10, 1989||Paradyne Corporation||Interleaved synchronous bus access protocol for a shared memory multi-processor system|
|US4821174||Jun 21, 1988||Apr 11, 1989||Westinghouse Electric Corp.||Signal processing system including a bus control module|
|US4941088||Feb 5, 1985||Jul 10, 1990||Digital Equipment Corporation||Split bus multiprocessing system with data transfer between main memory and caches using interleaving of sub-operations on sub-busses|
|US4953072||May 1, 1987||Aug 28, 1990||Digital Equipment Corporation||Node for servicing interrupt request messages on a pended bus|
|US5001625||Mar 24, 1988||Mar 19, 1991||Gould Inc.||Bus structure for overlapped data transfer|
|US5146597||Nov 19, 1991||Sep 8, 1992||Digital Equipment Corporation||Apparatus and method for servicing interrupts utilizing a pended bus|
|US5191649||Dec 21, 1990||Mar 2, 1993||Intel Corporation||Multiprocessor computer system with data bus and ordered and out-of-order split data transactions|
|US5235684||Jun 30, 1988||Aug 10, 1993||Wang Laboratories, Inc.||System bus having multiplexed command/id and data|
|US5237567||Oct 31, 1990||Aug 17, 1993||Control Data Systems, Inc.||Processor communication bus|
|US5274787||Jan 17, 1991||Dec 28, 1993||Nippon Telegraph & Telephone Corp.||Method of copy-back cache coherence control and tightly coupled multi-processor system with split transfer system bus|
|US5341495||Oct 4, 1991||Aug 23, 1994||Bull Hn Information Systems, Inc.||Bus controller having state machine for translating commands and controlling accesses from system bus to synchronous bus having different bus protocols|
|US5379384||Sep 23, 1993||Jan 3, 1995||Intel Corporation||Configuration data loopback in a bus bridge circuit|
|US5388224||Apr 24, 1992||Feb 7, 1995||Digital Equipment Corporation||Processor identification mechanism for a multiprocessor system|
|US5414820||Mar 21, 1994||May 9, 1995||Nexgen, Inc.||Crossing transfers for maximizing the effective bandwidth in a dual-bus architecture|
|US5483642||Sep 26, 1994||Jan 9, 1996||Hitachi, Ltd.||Bus system for use with information processing apparatus|
|US5594880||May 17, 1994||Jan 14, 1997||Motorola Inc.||System for executing a plurality of tasks within an instruction in different orders depending upon a conditional value|
|US6128688||Apr 23, 1999||Oct 3, 2000||Hitachi, Ltd.||Bus control system|
|US6219738 *||Feb 28, 2000||Apr 17, 2001||Hitachi, Ltd.||Information processing system|
|US6584538||Nov 16, 2001||Jun 24, 2003||Hitachi, Ltd.||Information processing system|
|JPH01161461A||Title not available|
|JPH03102558A||Title not available|
|JPH03222543A||Title not available|
|JPH03278156A||Title not available|
|1||"Futurebus+ P896.1: Logical Layer Specifications", IEEE, 1990, pp. 89-90.|
|2||J. Cantrell, "Futurebus+ Cache Coherence"; WESCON '89 Conference Record, Nov. 14-15, 1989, pp. 602-607.|
|3||J.A. Gallant, "Futurebus+", EDN, Oct. 1, 1990, pp. 87-98.|
|4||Langendoen et al, "Evaluation of Futurebus Hierarchical Caching", vol. 1, PARLE '91-Parallel Architectures and Language Europe, 1991, pp. 52-68.|
|5||M. Azimi et al, "Design and Analysis of a Hierarchical Snooping Cache Coherence System", Proceedings of the 27th Annual Allerton Conference on Communication, Control and Computing, vol. 1, 1988, pp. 109-118.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US20050165988 *||Dec 27, 2004||Jul 28, 2005||Sharp Kabushiki Kaisha||Bus communication system|
|U.S. Classification||710/309, 710/306, 710/305|
|International Classification||G06F13/36, G06F13/40, G06F13/00, G06F13/368, G06F5/00, G06F3/00, G06F13/14|
|Cooperative Classification||G06F13/4027, G06F13/36|
|European Classification||G06F13/36, G06F13/40D5|